Portal de Periódicos da CAPES

Corneal surface reconstruction using adult mesenchymal stem cells in experimental limbal stem cell deficiency in rabbits

2009; Wiley; Volume: 89; Issue: 8 Linguagem: Inglês

10.1111/j.1755-3768.2009.01812.x

1755-3768

Helga Reinshagen, Claudia Auw‐Haedrich, R. V. Sorg, Daniel Boehringer, Philipp Eberwein, Johannes Schwartzkopff, R. Sundmacher, Thomas Reinhard,

Corneal surgery and disorders

Purpose: To investigate the ability of mesenchymal stem cells (MSC) to transdifferentiate to corneal epithelial cells in experimental limbal stem cell deficiency in rabbits. Methods: Total limbal stem cell deficiency was produced in 21 right eyes of 21 New Zealand rabbits; 6 eyes served as controls (group 1, G1). After removal of the conjunctival overgrowth, five eyes received amniotic membrane transplantation (AMT; G2). In four eyes, autologous limbal stem cell transplantation from the healthy eye was performed with AMT (G3). In another six eyes, enriched autologous MSC were injected under the amniotic membrane (AM) (G4). Within 280 days, corneoscleral discs were analysed for goblet cells, cytokeratin (CK) 3/12, connexin 43, β1-integrin, CK 19, α-enolase, p63 and ATP-binding cassette transporter subtype G-2 (ABCG-2) distribution patterns. Results: Cultivated MSC were positive for CK 3/12 and α-enolase, but negative for ABCG-2, p63 and connexin 43. On rabbit corneas, CK 3/12 was expressed in all corneal regions in all groups, but with significantly different intensities. Among all other parameters, expression levels of ABCG-2, β1-integrin and connexin 43 were significantly different between the transplanted groups and the control group. After a mean follow-up time of 172 (47–280) days, goblet cells were rarely present in the central cornea (G1-4). Conclusion: CK 3/12 is not highly specific for differentiated corneal epithelium. Further, goblet cells are not a reliable marker for conjunctivalization in rabbits. Expression of ABCG-2, β1-integrin and connexin 43 after mesenchymal stem cell transplantation may indicate their ability to maintain their stem cell character or to transdifferentiate to epithelial progenitor cells. Corneal surface reconstruction is promoted by presumed stem cells, which are mainly located in the basal epithelial layers of the limbal region, the outer vascular rim at the junction between the cornea and conjunctiva (Kenyon & Tseng 1989; Kruse 1996; Schermer et al. 1986; Cotsarelis et al. 1989). Newer findings, nevertheless, seem to support the ability of centrally the limbal area located corneal epithelial cells to regenerate corneal epithelium (Majo et al. 2008). Severe limbal stem cell deficiencies lead to conjunctivalization and corneal opacification followed by neovascularization, recurrent ulcers and vision loss. In these cases, limbal tissue transplantation is necessary to restore eye function. If only one eye is affected, autologous limbal tissue can be obtained from the healthy eye and transplanted to the affected eye (Kenyon & Tseng 1989; Daya & Ilari 2001). The main advantage of this method is the lack of immune reactivity and therefore immunosuppression is not necessary. If the graft is too large, however, limbal stem cell deficiency can be induced in the healthy eye (Kenyon & Tseng 1989; Kruse & Reinhard 2001). In cases with bilateral limbal stem cell deficiency, homologous limbal grafts are necessary, with keratoplasty as a second step or homologous penetrating central limbo-keratoplasty in one step (Kenyon & Tseng 1989; Reinhard et al. 1999). Because of vascularization and resident antigen-presenting cells, immune reactivity is increased, and systemic immunosuppression is necessary for graft survival (Holland & Schwartz 2000; Reinhard et al. 1996; Reinhard & Sundmacher 1997; Sundmacher et al. 1992). Administered over a long period of time, immunosuppression carries the risk of side-effects such as disorders of liver and renal function, arterial hypertension, bone marrow depletion and neurotoxicity (Heering et al. 1991; Kutkuhn et al. 1997). Even with optimal immunosuppressive treatment, the mid-term clear graft survival ranges between 28% (Reinhard et al. 1999) and 70% (Holland & Schwartz 1996; Sundmacher & Reinhard 1996), depending on the stage of the disease. Postoperative results are improved by HLA matching, but that can lead to a longer waiting time for the corneal graft (Daya & Ilari 2001; Reinhard et al. 2004; Völker-Dieben et al. 2000). Amniotic membrane (AM) has healing properties in severe ocular surface disorders and is a useful substitute for the epithelial basement membrane (Kim & Tseng 1995; Lee & Tseng 1997; Kruse & Meller 2001). In partial limbal stem cell deficiencies, amniotic membrane transplantation (AMT) can be used to restore the corneal surface and, thus, visual acuity (Meller & Tseng 2000; Tseng et al. 1998). This surgical procedure alone, however, is not sufficient for severe limbal stem cell deficiencies, but improves the corneal situation in combination with limbal stem cell transplantation by reducing postoperative inflammation (Meller & Tseng 2000; Tseng et al. 1998). A more recent approach is the ex vivo expansion of limbal epithelial cells on AM (Tseng et al. 2000; Anderson et al. 2000). These epithelial sheets are transplanted to the eye and can restore the corneal surface for up to 2 years (Tseng et al. 2000; Pellegrini et al. 1997; Meller & Kruse 2001; Schwab et al. 2000). This method, however, is only successful in unilateral limbal stem cell deficiency. Homologous ex vivo expanded limbal stem cells are not superior to homologous limbal grafts (Cooper et al. 2002). Mesenchymal stem cells (MSC) have attracted attention as a better treatment option with less or no immunogenic potential. Adult human bone marrow MSC are able to differentiate into various cell lineages like human cardiac muscle cells in vivo (Strauer et al. 2001), human and mouse neural cells in vitro (Sanchez-Ramos et al. 2000) and rat limbus-like stem cells in vivo (Ma et al. 2006). Therefore, MSC may be a reliable source for epithelial progenitor cells. The expression of markers, such as ATP-binding cassette transporter subtype G-2 (ABCG-2), cytokeratin 19, α-enolase and β1-integrin, are characteristic of putative limbal stem cells (De Paiva et al. 2005; Wei et al. 1993; Zieske et al. 1992; Kreidberg 2000). The expression of cytokeratin 3/12 and connexin 43, however, would indicate differentiated corneal epithelial cells (Schlötzer-Schrehardt & Kruse 2005; Chen et al. 2004; Grueterich et al. 2002; Matic et al. 1997; Budak et al. 2005; Lavker et al. 2004). To examine our hypothesis that MSC have the potential to transdifferentiate into progenitor or differentiated epithelial corneal cells, we assessed the behaviour of bone-marrow stem cells in rabbit eyes with total limbal stem cell deficiency. The application for animal experiment was approved by the ethics committee of the Veterinary Institute of Heinrich-Heine-University, Düsseldorf, Germany. Care and treatment of animals were in accordance with institutional guidelines and with the World Medical Association Declaration of Helsinki and were all performed in the institute and under the institute’s supervision. Male New Zealand rabbits (2–3 kg body weight) were used as a model of ocular surface injury by chemical and mechanical removal of the epithelium and 360° limbectomy (Tsai et al. 1990). All treatments were performed under general anaesthesia. Depending on the time of surgery, anaesthesia was induced by intramuscular injection with fentanyl-fluanisone (Hypnorm®) and diazepam alone or together with intubation and isoflurane inhalation. After ocular surface injury, a total of 21 rabbits were distributed into four treatment groups (Table 1). Six eyes served as untreated controls (group 1). Groups 2, 3 and 4 had a second surgery at least 4 weeks later to remove the conjunctival overgrowth and cover the area with human AM. Briefly, the conjunctival overgrowth was removed with hockey and lamellar dissection knives. Preserved AM was used to cover the entire cornea and fixed tightly with 9-0 nylon sutures onto the episclera. The membrane rim was covered with conjunctiva, which was sutured with 8-0 Vicryl®. Group 3 received the same treatment as group 2 and had two pieces of autologous limbal tissue (6 mm× 2 mm in size, prepared from the healthy eye during the same surgery) transplanted to the injured eye at the upper circumference of the cornea before covering with AM. Group 4 received the same treatment as group 2 and had cultivated autologous MSC injected under the AM. The right eye of each animal was treated with n-heptanol to induce the epithelial abrasion: a corneal tip wetted with n-heptanol was rubbed over the corneal surface for 1–2 min. After rinsing with water, the remaining epithelium was removed with a hockey knife (Geuder, Heidelberg, Germany). Then, a partial-thickness corneal incision was made using a blade over 360° at 2 mm central from the limbus, followed by a deep lamellar dissection towards the limbus. The conjunctiva was dissected with scissors at 2 mm peripheral from the limbus. The lids were sutured with 5-0 Prolene®. At the end of the surgery, steroid and antibiotic ointment (Dexamytrex®; Mann, Berlin, Germany) was applied. For analgesia, nonsteroidal antiphlogistics (Carprofen®; Pfizer, St Louis, MO, USA) were given intramuscularly once per day for 1 week. After 3 days, the lid sutures were removed. With informed consent, a human placenta was donated after Caesarian section and freshly prepared under sterile conditions. As previously described, the AM was smoothly dissolved, cut in 2 cm× 2 cm pieces and fixed on pieces of cellulose with 6-0 Prolene® sutures (Kruse & Reinhard 2001). The samples were stored at −80°C in cell culture medium and glycerol in a 1:1 ratio (Kruse & Reinhard 2001). The group 4 rabbits underwent bone marrow aspiration to harvest the MSC. Under general anaesthesia, the intertrochanteric fossa of the left femur was prepared as described previously (Horan et al. 1980). The bone was penetrated at a 90° angle with a paediatric 18-gauge bone marrow punction needle. Bone marrow (BM) was aspirated to a volume of 1–2 ml in a syringe filled with 10 000 IU heparin in 1 ml. Connective tissue and skin was then sutured with soluble sutures. The BM samples were processed by diluting in 50 ml cold ammonium chloride solution to lyse erythrocytes. After incubation for 10 min at +4°C, samples were centrifuged for 7 min at 1500 rpm. After a washing step with phosphate-buffered saline (PBS), cells were resuspended in Dulbecco’s minimum essential medium, supplemented with 10% fetal bovine serum, 100 U/ml penicillin and 100 μg/ml streptomycin (all from Lenza, Verviers, Belgium) and cultivated at +37°C. At 1 week later, nonadherent cells were discarded, and the medium was exchanged. After this initial culture period, cells were passaged, when they reached 80% confluency. Mononuclear BM cells were identified microscopically. When the cell number reached 5 × 107, the cell cultures were harvested. Cells were then either processed for the next surgical procedure in the rabbit or stored at −80°C using 4% HSA/11% DMSO (human serum albumin; Octapharma, Langenfeld, Germany; dimethyl sulfoxide; Sigma-Aldrich, Taufkirchen, Germany) in RPMI-1640 (Biochrom, Berlin, Germany). The group 4 rabbits underwent the second surgical procedure on the injured eye 3–4 weeks after bone marrow aspiration (Table 1). After AMT and fixation, the 0.5-ml sample of MSC in buffered saline solution (BSS) was slowly and carefully injected underneath the membrane. The conjunctiva was tightly stretched over the membrane rim. All rabbits were examined every 3 days after the surgical procedure for 2 weeks, then weekly for 6 weeks, then monthly for up to 6 months. For examination, rabbits were sedated with ketamine (Ketanest®; Pfizer, St Louis, MO, USA) intramuscularly. A surgical microscope was used for precise examination. For clinical evaluation, corneal erosion (with fluorescein), the size and number of quadrants of corneal neovascularization were assessed. Two animals each in groups 2, 3 and 4 were killed for evaluation within 8 weeks. All other animals were killed and evaluated after 6–8 months. All injured eyes and 12 untreated left eyes were enucleated. Corneoscleral buttons with a 15-mm diameter were prepared. The specimens were fixed in 4% formaldehyde in 0.075 m phosphate buffer for 24 hr, dehydrated in increasing concentrations of ethanol (70–99%) and infiltrated with paraffin (Merck, Darmstadt, Germany) at 60°C. Sections (3-μm thick) were cut and floated on de-ionized water at 45°C, and single sections were mounted on Superfrost Plus glass slides (Menzel-Glaser, Braunschweig, Germany). Slides were subsequently dried at 60°C for 1 hr. The Periodic Acid Schiff-stained slides were examined to detect the presence of goblet cells. All corneal specimens and MSC were stained immunohistochemically according to the methods described in Table 2. Paraffin corneal and limbal sections were analysed each in three regions (central cornea, peripheral cornea and limbus) and in three epithelial layers (basal, suprabasal and superficial). The layers were judged semi-quantitatively regarding the intensity of their immunohistochemical staining (0 = no staining, 1 =weak, 2 = moderate, 3 = strong, 4 = very strong staining) and amount of cells stained positively (0 = negative, 1 ≤ 25%, 2 = 26–49%, 3 = 50–74%, 4 ≥ 75%) resulting in a total score calculated from the sum of both scores divided by 2 by a masked observer at a magnification of 200. Testing for interdependence of frequency data (immunohistochemical evaluation, amount of goblet cells) was carried out using Pearsons’s Chi-square test. Calculation of clinical parameters (size of vascularization, number of quadrants of vascularization and size of corneal erosion and their development) was carried out using the Kaplan–Meier estimator to assess median time. For both tests, a p-value of less than 0.05 (alpha error) was regarded as statistically significant. Table 1 summarizes the treatment groups. Immediately after injection of cultivated bone marrow stem cells underneath the AM, cell clusters could be observed under the membrane through the surgical microscope. At the first clinical visit three days later, clusters could not be detected anymore. Four to five weeks after injury, all eyes showed total conjunctivalization of the corneal surface (Fig. 1). The mean follow-up time after the whole surgical procedure was 172 days (range 47–280 days), which was the time-point at which the animals were killed and tissues harvested. In group 2, the cornea was covered with epithelium within a median of 28 days, which was proven by fluorescein staining. Vascularization occurred in all four quadrants within a median of 13 days and over the entire cornea within a median of 32 days. In group 3, the cornea was epithelialized within a median of 35 days and showed vascularization in all quadrants within a median of 22 days and over the entire cornea within a median of 45 days. In group 4, the corneal surface was epithelialized within a median of 34 days and showed vascularization in all quadrants within a median of 17 days and over the entire cornea within a median of 33 days. There were no statistically significant differences between groups. In summary, there were no differences in clinical appearance regarding corneal transparency, conjunctivalization and vascularization between all groups without change until the end of the observation period. Clinical appearance. (A) Complete limbal stem cell deficiency four weeks after inducing the abrasion; (B) four weeks after amniotic membrane transplantation (AMT) alone; (C) five weeks after AMT and autologous limbal stem cell transplantation underneath; (D) four weeks after AMT and Mesenchymal stem cells transplantation underneath. Microscopic and immunohistochemical evaluation was performed after the animals were killed. Table 3 gives details of the distribution of goblet cells. There were no statistically significant differences between groups. AM could not be identified in any case. Cultivated MSC were positive for cytokeratin (CK) 3/12 and α-enolase and negative for ABCG-2, p63 and connexin 43. Normal, healthy rabbit corneas showed mostly very strong expression of CK 3/12 in the central cornea, in the periphery and in the limbal area (Table 4). For connexin 43, mostly weak staining was detected in the central cornea, weak to no staining in the periphery and no staining in the limbal area (Table 4). All putative limbal stem cell markers were found in all corneal epithelial regions and not exclusively in the limbal area (Table 4). α-Enolase was detected from weak to very strong staining in equal proportions in the limbal area, in the periphery and even in the central cornea (Table 4). Normal rabbit conjunctiva was negative for CK 3/12. The level of CK 3/12 expression was significantly different between group 1 (strongest expression), and groups 3 and 4, in contrast to group 2 (lowest expression; Table 4, Fig. 2). Connexin 43 expression levels were significantly higher in the peripheral cornea in groups 3, 2 and 4, in contrast to group 1 (Table 4). ABCG-2 expression levels were significantly higher in the central cornea in groups 3 and 4 compared to groups 1 2 (Table 4, Fig. 2). β1-integrin expression levels were significantly higher in the central cornea in groups 3 and 4 compared to the control group, in contrast to group 2 (Table 4, Fig. 2). Expression levels of CK 19, α-enolase and p63 did not differ significantly between groups. There was no significant difference between groups with a shorter ( 60 days) follow-up time. Immunohistological appearance. (A) positive ATP-binding cassette transporter subtype G-2 (blue granular cytoplasmic) combined with p63 (brown nuclear) staining with most intense positive staining of the basal cell layer in the central part of a cornea of group 4; (B) positive staining for β1-integrin (blue granular cytoplasmic) in the centre of a cornea of group 4; (C) positive staining for cytokeratin (CK) 3/12 (red cytoplasmic) in the centre of a cornea of the control group with increasing intensity from basal towards superficial layer; (D) positive staining for CK 3/12 (red cytoplasmic) in the centre of a cornea of group 4 also with increasing intensity from basal towards superficial layer. MSC are increasingly important as a multipotent autologous cell source. There are encouraging reports demonstrating the plasticity of adult MSC that are directly placed on damaged tissue i.e. cardiac muscle cells (Strauer et al. 2001). Ma et al. transplanted ex vivo expanded human MSC on human AM on rat corneas that had been injured 7 days before: they concluded that the therapeutic effect was a result of the inhibition of inflammation and angiogenesis rather than epithelial differentiation (Ma et al. 2006). The authors concluded that both the AM and expanded MSC contributed to surface reconstruction (Ma et al. 2006). In the present study, we examined the behaviour of MSC directly placed on the corneal stroma underneath AM. Immediately after injection cell clusters were observed. Three days later, the clusters could not be detected anymore. We can give no evidence, if these cells stayed in place. There are three possibilities: first, cells were spread evenly on the cornea; second, cells spread also underneath the conjunctiva; third, cells were washed out by the tear film. The intention to use AM was, first, to create some kind of reservoir for the suspended MSC by fixing the AM as tightly as possible, and, second, to use special properties of the AM. As shown for ex vivo expansion of human limbal stem cells, AM has the ability to maintain stem or progenitor cell characteristics, even better after denudation of the devitalized amniotic epithelium (Grueterich et al. 2002). The use of human AM in such a limbal deficiency rabbit model as chosen herein has been described before. There seems to be no need to use rabbit AM, because an acellular membrane is transplanted and therefore immunoreactive processes are avoided (Kim & Tseng 1995; Koizumi et al. 2000; Wang et al. 2003). Moreover, human AM also maintains rabbit limbal epithelial cell sheet outgrowth from a limbal explant (Wang et al. 2003). The enucleation specimens of groups 2 and 4 did not show AM microsopically. This is consistent with previous observations that AM is absorbed within 5 weeks in rabbits (Du et al. 2003). After AMT, which was performed in all treatment groups, increased vascularization in all quadrants with rapid conjunctival epithelial repair of the surface and decreased corneal transparency were observed uniformly, even in group 3, where at least partially restored corneal transparency had been expected. We assume that the use of AM in this group might have ruined the effect of autologous limbal tissue. AM induces epithelial repair with dense vascularization. This is confirmed by the literature, where this phenomenon was described previously (Kim & Tseng 1995; Ti et al. 2002). Time to enucleation for examination in our study was a minimum of 47 days. Microscopically, goblet cells were virtually absent on the corneal surfaces. These findings confirm former clinical results that the conjunctival repair of a damaged corneal surface in rabbits by 4 weeks post wounding showed an epithelium histologically similar to normal corneal epithelium (Wei et al. 1993). Goblet cells, which migrated first onto the corneal surface, disappeared within a few weeks (Wei et al. 1993). Therefore, the presence of goblet cells is not a reliable marker of conjunctivalization, as in humans, as have been shown in other studies (Auw-Haedrich et al. 2009). Moreover, Wei detected small amounts of CK 3 and 12 expression after corneal wounding alone in the covering conjunctival epithelium in rabbits (Wei et al. 1993). Also in human conjunctival epithelium, Kawasaki detected clusters of CK-12-positive cells, appearing to be ectopically residing corneal epithelial cells (Kawasaki et al. 2006). A third working group found a covering epithelium displaying a corneo-epithelial phenotype with full-thickness CK 3/12 expression in a rabbit limbal deficiency model, similar to that used herein (Kim and Tseng 1995). These findings support our observations of CK 3/12 expression in all treatment groups. The highest level was observed in group 1 as term of the special repair mechanism in rabbits as described previously. The difference to groups 2, 3 and 4 may be the second surgical procedure, in which the first conjunctival cover had to be removed, and the cornea has been covered with AM. Rabbit MSC are positive for CK 3/12, which has not been described before to our knowledge. This might be a given fact or be explained by a cross-reaction of the anti-human CK 3/12 antibody. In accordance, one would have expected the highest expression level in group 4, but this was not the case. Even normal rabbit corneas showed an unexpected expression of CK 3/12 in the limbal area. Together with the literature, these are all indications that CK 3/12 is not specific for the differentiated corneal epithelium of epithelial stem cell origin in rabbits. Since the normal rabbit conjunctiva was negative for this marker, this might indicate that ‘insufficient’ corneal epithelium and/or transdifferentiated MSC were stained positively. Connexin 43 is a molecule that mediates gap junction intercellular communication and indicates epithelial differentiation (Matic et al. 1997). It is absent in human limbal stem cells, but is expressed in very low levels in the rabbit limbus (Matic et al. 1997). In this study, it was detected in the periphery of the cornea significantly more often in groups 3 and 2, followed by the group 4. An interpretation might be some kind of differentiation of epithelial cells spread out to the peripheral cornea. In normal rabbit corneas, there was weak staining in the central cornea, weak to no staining in the periphery and no staining in the limbal area. Immunohistochemical analysis indicated that ABCG-2, a molecular determinant of human MSC and marker of clonogenic human and rabbit limbal epithelial cells (De Paiva et al. 2005; Budak et al. 2005), was present in the central cornea at significantly higher levels in group 4, followed by group 3. This is a result, which we theoretically have expected. But, in normal rabbit corneas, there was no higher expression of ABCG-2 in the limbal area, and the rabbit MSC were negative for ABCG-2. The presence of ABCG-2 positive cells in group 1 – nevertheless at a lower level compared to the other groups – might indicate that some of the cells expressing this marker escaped the limbo-destructive treatment, or centrally located corneal stem cells (Majo et al. 2008) were stained positively here. β1-Integrin, an extracellular matrix component, is expressed in the basal cells of the human limbus (Kreidberg 2000; Wang et al. 2003), but also in the rabbit basal limbal cells (Espana et al. 2003). In our study, β1-integrin was detected in the central cornea in a statistically significant higher intensity in groups 3 and 4. Despite small numbers, this might confirm our theoretical considerations. In normal rabbit corneas, β1-integrin was mostly not detected, even in the limbal area. There were no significant differences in staining for α-enolase, a protein with increased concentrations in human limbal epithelial cells (Zieske et al. 1992); CK 19, a human and rabbit limbal cytokeratin (Wei et al. 1993); and transcription factor p63, characterizing immature human and rabbit epithelial cells (Du et al. 2003). In normal rabbit corneas, all these three markers showed an unexpected staining behaviour with α-enolase being detected even in the central cornea and in the periphery, and with both, CK 19 and p63, showing no striking staining in the limbal area. In summary, these findings might suggest that MSC in this context might be able to maintain their stem cell-like character or to transdifferentiate to putative epithelial progenitor cells with regard to the results of ABCG-2, β1-integrin and connexin 43. The differentiation marker CK 3/12 failed to show some evidence towards differentiation into epithelial corneal cells. Connexin 43 was the only reliable marker with regard to the results in normal rabbit corneas in our study. All other parameters did not clearly show the expected results, which limits the interpretation of results of the treated eyes. In our study, CK 3/12 positivism was not a proof for differentiation towards corneal epithelial cells, as also ‘insufficient’ epithelial cells stained positively with this marker. The findings of the present study might be a first step towards the utilization of MSC as an autologous source for corneal epithelial regeneration without immunogenic risk. Further studies are needed to determine the method for delivering enriched MSC to put on the denuded stroma, to keep them in place and to precisely characterize if and how MSC transdifferentiate into epithelial (progenitor) cells. Antiangiogenic agents might be necessary to suppress accompanying vascularization. We thank Mrs. Gerlinde Westphal and Mrs. Renate Buchen for excellent technical assistance.